Fast Method to Search for Linear Frequency-Modulated Signals

1. A method for determining frequency and slope of a linear frequency modulated signal for implementation on a computer comprising the steps of: collecting windowed data samples concerning the signal; formulating a search surface function from said windowed data samples; dividing the search surface function into a number of parts; calculating the search surface function values at a selected number of frequency and slope values, said selected number of frequency and slope values being related to the number of parts; calculating search surface function parts for each search surface data part at the selected frequency and slope values; interpolating the search surface function parts at values between selected frequency values for a selected slope to obtain first interpolated values, each interpolation between selected frequency values being an interpolation geometry; calculating a correlation coefficient for each interpolation geometry; determining the best interpolation geometry for the search surface function part at slopes between selected frequencies and selected slopes as the interpolation geometry that gives the highest correlation coefficient; interpolating to obtain search surface function part values at frequencies and slopes between selected frequencies and slopes according to the determined best interpolation geometry to obtain second interpolated values; adding the search surface function parts together to give a restored search surface function, said search surface parts having first interpolated values and second interpolated values incorporated therein; searching through frequency and slope values of said search surface function calculated values, first interpolated values, and second interpolated values to obtain the frequency and slope values that maximize the search surface function; and providing the frequency and slope values that maximize the search surface function as the frequency and slope of the signal.

2. The method of claim 1 wherein frequency and slope values crossing the Nyquist rate are omitted from the calculated search surface function part.

3. The method of claim 2 wherein frequency and slope values crossing zero are omitted from the calculated search surface function part.

4. The method of claim 3 further comprising the steps of: determining a threshold for the search surface function part; and wherein said step of interpolating the search surface function part at values between frequency values is only performed when the search surface function value at the frequency value exceeds said determined threshold; said step of interpolating to obtain search surface function part values at frequencies and slopes between calculated frequencies and slopes is only performed when the search surface function part value at one of the calculated frequency values and slopes exceeds said determined threshold.

5. The method of claim 1 wherein the steps of dividing the search surface function values, calculating the search surface function parts, interpolating to obtain first interpolated values, calculating a correlation coefficient, determining the best interpolation geometry, and interpolating to obtain second interpolated values are performed recursively until the search surface part contains a preselected number of windowed data samples.

6. The method of claim 5 wherein frequency and slope values crossing the Nyquist rate are omitted from the calculated search surface function part.

7. The method of claim 6 wherein frequency and slope values crossing zero are omitted from the calculated search surface function part.

8. The method of claim 7 further comprising the steps of: determining a threshold for the search surface function part; and wherein said step of interpolating the search surface function part at values between frequency values is only performed when the search surface function value at the frequency value exceeds said determined threshold; said step of interpolating to obtain search surface function part values at frequencies and slopes between calculated frequencies and slopes is only performed when the search surface function part value at one of the calculated frequency values and slopes exceeds said determined threshold.